A background-free direction-sensitive neutron detector
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Roccaro, Alvaro et al. “A Background-free Direction-sensitive
Neutron Detector.” Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and
Associated Equipment 608.2 (2009): 305–309.
As Published
http://dx.doi.org/10.1016/j.nima.2009.06.102
Publisher
Elsevier
Version
Author's final manuscript
Accessed
Wed May 25 15:17:57 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/76319
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike 3.0
Detailed Terms
http://creativecommons.org/licenses/by-nc-sa/3.0/
1
1
A Background-Free Direction-Sensitive Neutron
2
Detector
3
Alvaro Roccaro,1* H. Tomita,1 S. Ahlen,1 D. Avery,1 A. Inglis,1 J. Battat,2 D.
4
Dujmic,2 P. Fisher,2,4,5 S. Henderson,2 A. Kaboth,2 G. Kohse,2 R. Lanza,2 J.
5
Monroe,2 G. Sciolla,2 N. Skvorodnev,3 H. Wellenstein,3 and R. Yamamoto2
6
1
Physics Department, Boston University, Boston, MA 02215, 2Physics Department and
7
Laboratory for Nuclear Science, Massachusetts Institute of Technology, Cambridge,
8
MA 02139, 3Physics Department, Brandeis University, Waltham, MA 02454, 4MIT Kavli
9
Institute and Institute for Soldier Nanotechnology, Massachusetts Institute of
10
Technology, Cambridge MA 02139, 5MIT Kavli Institute, Massachusetts Institute of
11
Technology, Cambridge MA 02139 *Deceased
12
We show data from a new type of detector that can be used to determine neutron
13
flux, energy distribution, and direction of neutron motion for both fast and
14
thermal neutrons. Many neutron detectors are plagued by large backgrounds from
15
x-rays and gamma rays, and most current neutron detectors lack single-event
16
energy sensitivity or any information on neutron directionality. Even the best
17
detectors are limited by cosmic ray neutron backgrounds. All applications
18
(neutron scattering and radiography, measurements of solar and cosmic ray
19
neutron flux, measurements of neutron interaction cross sections, monitoring of
20
neutrons at nuclear facilities, oil exploration, and searches for fissile weapons of
21
mass destruction) will benefit from the improved neutron detection sensitivity and
22
improved measurements of neutron properties made possible by this detector. The
23
detector is free of backgrounds from x-rays, gamma rays, beta particles,
24
relativistic singely charged particles and cosmic ray neutrons. It is sensitive to
25
thermal neutrons, fission neutrons, and high energy neutrons, with detection
26
features distinctive for each energy range. It is capable of determining the location
2
1
of a source of fission neutrons based on characteristics of elastic scattering of
2
neutrons by helium nuclei. The detector we have constructed could identify one
3
gram of reactor grade plutonium, one meter away, with less than one minute of
4
observation time.
5
We report a significant advance in the detection of neutrons and determination of
6
their energy and direction of motion in a simple, compact device. The concept [1-3] is
7
illustrated in Figure 1. The detector is in a chamber that contains CF4 gas at low
8
pressure. One bar or more of 4He is added to provide a target for neutrons through
9
elastic scattering, and a few torr of 3He is added to detect thermal neutrons via the
10
reaction n+3He→3H+p. The relative amounts reflect the 8b cross section for elastic
11
scattering of neutrons on 4He and the 5327 barn cross section for the absorption of
12
thermal neutrons by 3He. A cathode mesh and field cage set up an electric field in an
13
electron drift region, and a grounded mesh is separated by a few hundred microns from
14
an anode copper plate that sets up a high electric field in an amplification region. A
15
charged particle moving through the chamber leaves a trail of ionization electrons in its
16
wake. The electrons from this track drift to the amplification plane, where an electron
17
avalanche occurs, accompanied by the emission of scintillation light from CF4. The
18
scintillation light is imaged by the lens1 and CCD camera2. If a neutron collides
1
We use a Nikon lens with a focal length of 55 mm and an f number of 1.2. For
the alpha particle images of Figure 2, the lens was 39 cm from the amplification mesh.
For the images of Figure 3, the lens was 28 cm from the amplification mesh.
2
The Apogee U6 camera used a Kodak KAF-1001E CCD chip, which has a 1024
x 1024 array of 24 x 24 micron pixels, and has a read noise of 7 electrons. The quantum
efficiency is 65% at 630 nm (the peak of the CF4 emission spectrum2). We used 4 x 4
hardware binning for the pixels.
3
1
elastically with a 4He nucleus it produces a track similar to that of an alpha particle from
2
a radioactive source, except at lower energy. If a neutron is absorbed by a 3He nucleus it
3
produces back-to-back tracks of a proton and triton with 764 keV of kinetic energy. The
4
device is blind to minimum ionizing charged particle tracks and has image formation
5
and clearing time of the order of 10 ns. Determination of the third component of tracks
6
can be made with the use of a photomultiplier tube to monitor drift time, which is up to
7
3 µs.
8
Our concept is similar to, but much simpler than, one proposed [4] in 1994 to
9
search for dark matter. That device utilized a time projection chamber with parallel
10
plate avalanche counter, an optical readout using the light-emitting gas triethylamine
11
(with 280 nm wavelength), and a 4.5 kG magnetic field to reduce electron diffusion.
12
The light collecting system included an ultraviolet lens and an image intensifier that was
13
followed by a phosphor screen. The image on the phosphor screen was viewed by a
14
second lens and a CCD video camera. Another device similar to ours has been built [5]
15
to detect thermal neutrons using a Gas Electron Multiplier (GEM) in a 5 cm diameter
16
chamber with a 2 cm drift distance. The gas used was a mixture of CF4 (300 torr) and
17
3
18
thin to use with low pressure CF4, which is why we adopted the mesh approach. In
19
addition, GEMs are quite expensive and fragile in comparison to the meshes that we
20
use.
He (460 torr). Our early work showed that commercial 50 µm thick GEMs were too
21
Other standard neutron detectors currently in use include the 3He based Bonner
22
sphere technique [6], pulse shape discrimination in organic scintillators to distinguish
23
elastic collisions of neutrons with protons from gamma ray signals [7-8], and various
24
schemes making use of other exoergic thermal neutron capture reactions (including e.g.
4
1
n + 6 Li → α + 3 H + 4.786 MeV , or n +10 B → α + 7 Li + 2.79 MeV ). For fast neutrons,
2
our imaging technique is superior because no moderator is needed and the image of the
3
recoil track gives both energy and direction information about the incident neutron. For
4
thermal neutrons, our imaging method gives the interaction location within the detector
5
with 200 µ precision and is insensitive to gamma and beta rays, both of which are
6
important for background rejection.
We recently exposed a prototype detector to several sources of radiation,
7
8
including an 241Am alpha source, a 252Cf fission spectrum neutron source3, a deuterium-
9
tritium neutron generator4, and cosmic ray neutrons. We used an Apogee U6 CCD
10
camera and a Nikon 55 mm lens, which provided a field of view 10 cm on a side for the
11
neutron exposures. The field cage drift was 10 cm high, and the meshes that bounded
12
the drift region were made of stainless steel. Figure 2 shows a pair of tracks from the
13
alpha source. The field of view was 14 cm for this setup. The light intensity is indicated
14
by the color scale, and the Bragg peak of ionization is apparent. The gap in one of the
15
tracks is due to a nylon wire that separates the mesh from the copper anode. The
16
direction of motion of a stopping alpha particle can be determined by the reduction of
17
ionization as the alpha particle approaches its end of range.
Figure 3 shows 15 successive recoil events from collisions with 252Cf neutrons.
18
19
The amplification plane was parallel to the floor of the lab and the source was six
20
meters away, about the same height from the floor as the detector. The neutrons came
21
from the bottom of the images. This can be observed from the orientation of the recoils
3
The
252Cf
source
emitted
12
million
neutrons/s.
Approximate
cosmic
ray
neutron
rates8
are
10
thermal
neutrons/m2/s,
20
neutrons/m2/s
for
neutrons
with
energy
in
the
MeV
range,
and
20
neutrons/m2/s
for
neutrons
with
energy
>
20
MeV.
4
The
Thermo
MF
Physics
A‐325
neutron
generator
produced
14.1
MeV
neutrons
at
the
rate
of
50
million
neutrons/s.
5
1
and the diminishing of light intensity toward the end of range. One of the tracks is going
2
the wrong direction, and is most likely a neutron that scattered off the wall of the lab.
3
The excellent directionality information is due to the fortuitous circumstance that the
4
differential elastic cross section for n –4He scattering in the MeV region is largest for
5
center-of-mass scattering angles of 180 degrees. Each of the images of Figure 3 had a
6
one-second exposure time, and about one out of every three exposures had a neutron
7
recoil event. The rate of recoil events from cosmic rays was observed to be about one
8
per hour prior to the introduction of the neutron source.
9
We have developed image analysis software to automate data analysis. Tracks are
10
found and measured and scatter plots of projected track length vs. total energy are
11
obtained5. This allows different particle types to be identified. Improved resolution will
12
be possible with the use of photomultiplier tube information, which enables the
13
determination of total track length. Figure 4 shows scatter plots for three cases: (a) 80
14
torr CF4 (Vamp = 700 volts, Vdrift = 250 volts/cm), 17 hr exposure to cosmic rays; (b) 80
15
torr + 1 torr 3He, 18.5 hr exposure to 252Cf + 5 hr to cosmic rays (Vamp = 750 volts, Vdrift
16
= 250 volts/cm); (c) 80 torr CF4 + 560 torr 4He, 18.5 hr exposure to 252Cf + 5 hr to
17
cosmic rays (Vamp = 750 volts, Vdrift = 250 volts/cm) and (d) 80 torr CF4 + 560 torr 4He
18
+ 1 torr 3He,18.5 hr exposure to 252Cf + 5 hr to cosmic rays (Vamp = 750 volts, Vdrift =
5
Track-finding for the image analysis includes background subtraction to
eliminate hot pixels, flat field corrections, median filters, thresholding and the
application of Hough transforms to search for lines in images. Once tracks are found
and thier0 lengths are determined from the Hough transform information, their pixel-bypixel light levels are determined, and the total light level, which is proportional to
energy loss, is measured.
6
1
250 volts/cm). The energy scales were determined from calibrations with alpha source
2
particle tracks.
3
In Figure 4d we see two bands, the upper being due to neutron 4He recoils, and the
4
lower being due to the triton + proton tracks from the n + 3He capture reaction. These
5
tracks are only partially contained in the volume of the detector so its energy is not
6
fixed at 764 keV. The elastic cross section for n + 4He elastic scattering is 8 barns at 1
7
MeV, and the thermal neutron capture cross section on 3He is 5327 barns. Thus, the
8
detection efficiencies are about the same for fission and thermal neutrons and the
9
ambient thermal neutron flux in the lab is found to be about the same as the fission
10
neutron flux. Figure 4a shows background events. These are of four main types. (i) The
11
left side of the plot has a population of very short tracks with large energy which are
12
due to background radiation interacting directly with the CCD. We have found by visual
13
inspection that these can be eliminated by track morphology cuts. They may also be
14
eliminated by using a photomultiplier tube or electrical readout of the amplification
15
plane in coincidence with the CCD exposure. (ii) most of the plot is populated by a
16
broad swath of events with ranges of a cm or more, and energies of 1/2 MeV or more.
17
These are due to alpha particles from radioactive nuclei in the decay chains of U/Th
18
contaminants found in chamber materials. Most of these are coming from the field cage
19
and visual inspection indicates they may be removed by rejecting recoil events that have
20
part of their track near the edge of the active region of the detector. Reduced
21
backgrounds can be obtained with the use of copper for the field cage and mesh instead
22
of stainless steel. (iii) there is a cluster of events with several mm of projected range and
23
less than 100 keV of energy. These are recoil nuclei from elastic scattering of cosmic
24
neutrons with 4He, as can be verified by examining Figure 4d which has no such
25
feature. These events may be removed with a small loss in efficiency by a minimum
26
projected range cut at 5 mm as may be seen by comparing Fig. 4a with Fig. 4c. (iv) the
27
events at the bottom of Figure 4a are due to x-rays which are just barely detectable in
7
1
pure CF4 which has a gain about twice that of the helium mixtures. Owing to their low
2
specific energy loss, these events are below threshold in a CF4-helium mixture. We
3
have developed algorithms to remove these backgrounds that will be reported in a
4
forthcoming publication.
5
Figure 5 shows the distribution azimuthal angles of helium nuclei recoils from an
6
exposure to the 252Cf source located at 90o with respect to the detector. The large peak
7
at 90 degrees corresponds to recoils caused by neutrons coming from that direction.
8
The peak near 270 degrees result from neutron that interact in the detector after
9
recoiling from the wall immediately behind. Relative sizes of the two peaks agree with
10
an estimated 13% of neutrons being reflected from the back wall. The smaller peaks in
11
the picture are consistent with neutron reflections from the side walls of the shielded
12
room.
13
We have built a large detector with a 20 cm drift field cage and a 30 cm diameter
14
amplification plane (using copper mesh), which was viewed by a Schneider 17 mm
15
focal length lens and an Andor iXon 888 EMCCD. This type of CCD has on-chip
16
amplification which enables single photon per pixel imaging. Figure 6 shows two
17
images from this device, with a gas mixture of 40 torr CF4 + 600 torr 4He, 620 volts on
18
the anode and a drift field of 125 V/cm. The one-second duration images were taken
19
with the Cf source at six meters distance. Neutrons came from the top of the images.
20
With a gas pressure of 4 bar, this same device would produce a similar image as these
21
for a one-minute exposure time if one gram of reactor grade plutonium were one meter
22
away. Figure 6 shows an image from the EMCCD setup for 14.1 MeV neutrons from
23
the neutron generator. A number of inelastic interactions become possible at these
24
higher energies, and identifying these interactions enables measurements of the high
25
energy neutron environment. The central event shows three alpha particles emerging
8
1
from the inelastic collision of a neutron with a carbon nucleus. The neutron energy
2
threshold for this reaction is 7.9 MeV.
3
We are currently designing an EMCCD6 based portable neutron detector with a
4
detection volume of 20 liters that will operate with a mixture of CF4, 4He, and 3He at a
5
pressure of 4 bars. The device, sketched in Figure 8, will be built and field tested for
6
several of the applications mentioned above. The techniques we have described here are
7
made possible by substantial improvements in CCD technology over the past decade. It
8
is likely that further developments will continue to be made in the next several years
9
that will lead to improved performance and to reduced costs for this new imaging
10
technology. Benefits to many fields other than neutron detection will probably become
11
possible.
12
Much of this work was carried out by Alvaro Roccaro, who passed away
13
unexpectedly during preparation of this letter. He will be sorely missed and we
14
dedicate this letter to his memory.
15
References
16
17
1. D. Dujmic et al., Nuclear Instruments and Methods In Physics Research A 584,
327 (2008).
6
The optics for the large detector include a Schneider 17 mm focal length lens
with f number of 0.95 and an Andor iXon DU-888 EMCCD. The EMCCD camera
utilizes an e2V CCD201-20 chip. The chip is back-illuminated with a quantum
efficiency of 95% at 630 nm, and has a 1024 x 1024 array of 13 x 13 micron pixels. We
used 2 x 2 hardware binning for the pixels.
9
1
2
2. A. Kaboth et al., Nuclear Instruments and Methods In Physics Research A 592,
63 (2008).
3
3. D. Dujmic et al., Astroparticle Physics 30, 58 (2008).
4
5
4. K.N. Buckland, M.J. Lehner, G.E. Masek and M.Mojaver, Physical Review
Letters 73, 1067 (1994).
6
7
5. F.A.F. Fraga et al., Nuclear Instruments and Methods In Physics Research A
478, 357 (2002).
8
9
6. H. Matsumoto, T. Goka, K. Koga, S. Iwa, T. Uehara, O. Sato, and S. Takgi,
Radiation Measurements 33, 321 (2001).
10
11
7. W.L. Bryan, C.L. Britton, J.T. Mihalczo, J.S. Neal, S.A. Pozzi, and R.W.
Tucker, Nuclear Science Symposium Conference Record 2, 1192 (2003).
12
13
8. Nakamura, T., Nunomiya, T., Abe, S., Terunuma, K., & Suzuki, H., Journal of
Nuclear Science and Technology 42, 843-853 (2005).
14
Acknowledgements We acknowledge support of this work by grants from the National
15
Science Foundation Department of Energy), Department of Homeland Security, the
16
MIT Physics Department, the MIT Kavli Institute and the Institute for Soldier
17
Nanotechnology at MIT.
18
Author Information Correspondence and requests for materials should be addressed to
19
Peter Fisher (fisherp@mit.edu).